fashion. Briefly, 18S sequence was PCR amplified by a single round RT-PCR ..... possible to predict disease relapse or cure only on the basis of a positive or ...
Leukemia (2001) 15, 1072–1080 2001 Nature Publishing Group All rights reserved 0887-6924/01 $15.00 www.nature.com/leu
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR in patients with INV(16) acute myeloid leukemia G Marcucci1, MA Caligiuri1, H Do¨hner2, KJ Archer1, RF Schlenk2, K Do¨hner2, EA Maghraby1 and CD Bloomfield1 1
Division of Hematology-Oncology, Department of Internal Medicine, Comprehensive Cancer Center, The Ohio State University, Columbus, OH, USA; and 2Division of Hematology-Oncology, University of Ulm, Germany
Amplification of the CBF/MYH11 fusion transcript by a qualitative reverse transcription-polymerase chain reaction (RTPCR) has been used to detect minimal residual disease (MRD) and assess the risk for disease relapse in inv(16)(p13q22) acute myeloid leukemia (AML). This strategy has, however, produced conflicting results and because of an uncertain predictive value, its use in the clinical setting cannot be recommended. The objective of the current study was to evaluate if quantification by Real Time RT-PCR could be useful to determine levels of CBF/MYH11 fusion transcripts predictive of clinical outcome in inv(16)(p13q22) AML at diagnosis or during remission. Bone marrow (BM) samples from 16 patients with inv(16) AML enrolled on a German multicenter trial (AML HD93) were analyzed for levels of CBF/MYH11 fusion transcripts by Real Time RT-PCR at diagnosis (n = 14), during remission (n = 10) and at relapse (n = 6). The CBF/MYH11 transcript copy number in each sample was normalized to copies of an internal control housekeeping transcript (ie 18S). The copy number measured at diagnosis or relapse were 3 to 4 log higher that those measured during remission, following completion of induction treatment. A high CBF/MYH11 transcript copy number at diagnosis had a significant correlation with a high percentage of BM blasts (Spearman’s coefficient = −0.66; P = 0.03), and a borderline correlation with a short complete remission (CR) duration (Spearman’s coefficient = −0.51; P = 0.07). No difference in levels of CBF/MYH11 fusion transcripts measured during intensification therapy was found between patients destined to relapse and those who continued in CCR (P = 0.75). Following completion of the entire chemotherapy program, patients that during CR showed a CBF/MYH11 fusion transcript copy number ⬎10 had a significantly shorter CR duration (P = 0.002) and higher risk for disease relapse (P = 0.05) than patients with a CBF/MYH11 fusion transcript copy number ⬍10. The results of the current study, therefore, suggest that it is possible to determine in remission samples a threshold of CBF/MYH11 transcript copy number above which relapse occurs and below which continuous CR is likely. Leukemia (2001) 15, 1072–1080. Keywords: minimal residual disease; AML; inv(16); Real Time RTPCR; CBF/MYH11
Introduction Karyotype is one of the most important independent prognostic factors in acute myeloid leukemia (AML). In the majority of studies of adult primary AML, the presence of chromosomal abnormalities involving genes encoding for the ␣ or  subunits of core binding factor (CBF), t(8;21)(q22;q22) or inv(16)(p13q22), respectively, is associated with a very high complete remission (CR) rate (苲90%) and probability (50% to 70%) of remaining in CR at 5 years.1 A recent Cancer and Leukemia Group B (CALGB) analysis has indicated that these particular subsets of AML benefit the most from intensification treatment with high-dose cytarabine (HiDAC), supporting the Correspondence: G Marcucci, The Ohio State University, The Comprehensive Cancer Center, A433B Starling-Loving Hall, 320 West 10th Avenue, Columbus OH 43210, USA; Fax: 614–293–7527 Received 13 December 2000; accepted 5 March 2001
notion that cytogenetic analysis should be used to stratify AML patients into risk-adapted therapeutic subgroups.2 However, a significant number of patients with inv(16)(p13q22) AML relapse and die of their disease, despite optimal treatment. Therefore, it may be important to identify, within this cytogenetic subgroup, those patients at higher risk for disease relapse and assess whether diversified and more aggressive therapeutic approaches such as hematopoietic stem cell transplantation implemented before clinically detectable relapse, could allow them to achieve a long-term continuous CR (CCR). One of the strategies being explored to predict clinical outcome in AML patients with specific genomic rearrangements is detection of minimal residual disease (MRD) following achievement of hematologic remission. Since relapse is likely to occur as the result of treatment failure to completely eradicate leukemic blasts, detection of chimeric genes and their fusion transcripts using sensitive molecular methodologies has been utilized, during CR, as a surrogate marker for resistant disease.3 At the molecular level, inv(16)(p13q22) results in the fusion of the CBF gene at chromosome band 16q22 with the MYH11 gene at chromosome band 16p13, creating a new chimeric gene, CBF/MYH11.4 Since the genomic breakpoints within the CBF and MYH11 genes are variable, at least eight different types of CBF/MYH11 fusion transcripts (ie A to H) are encoded.5 The most common of these fusion transcripts is referred to as type A and is detected in approximately 85% of patients with AML and inv(16)(p13q22). Amplification of CBF/MYH11 fusion transcripts by a qualitative reverse transcription-polymerase chain reaction (RT-PCR) in bone marrow (BM) and blood (PB) samples has been used to detect MRD and assess the risk of relapse simply on the basis of a positive or negative RT-PCR assay in patients with inv(16)(p13q22) AML.6–12 This strategy has, however, produced conflicting results and, because of an uncertain predictive value, its use in the clinical setting cannot be recommended.12 More recently, two studies have suggested that quantification of CBF/MYH11 fusion transcripts during remission may prove to be more useful in predicting clinical outcome than a simple qualitative assay.13,14 We have previously reported the feasibility of rapid and accurate quantification of chimeric fusion transcripts in AML patients using the Real Time RT-PCR, a novel fluorometric PCR methodology that allows calculation of levels of specific targets during an ongoing PCR amplification, thus the name ‘Real Time’, and in a large dynamic range of absolute quantification.15 In the current study, we have used this methodology to quantitate the copy numbers of CBF/MYH11 fusion transcripts in patients with inv(16)(p13q22) AML at diagnosis, at different time-points during remission and at relapse. The major objective of this retrospective analysis was to determine if levels of CBF/MYH11 expression, at diagnosis and during remission, are potentially predictive of relapse or CCR.
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
Patients and methods
Patient samples Sixteen (Unique Patient Number (UPN) 1–16) of the 23 patients with inv(16)(p13q22) AML enrolled on the German multicenter AML HD93 protocol were studied at diagnosis and at different time-points following achievement of CR. Cytogenetic analysis for all samples from patients entered on AML HD93 was performed at a central laboratory using both G-banding analysis as well as fluorescence in situ hybridization (FISH). In the present study, the only criteria used to include or exclude patients were the availability for molecular analysis of sequential bone marrow (BM) samples. Selected clinical and molecular characteristics of each patient are reported in Table 1. The AML HD93 protocol was open for patient recruitment from July 1993 to January 1998 and karyotype was used to stratify post-remission therapy in adult patients with primary AML. Of the 16 patients analyzed in the current study, 15 received double induction therapy with Table 1
Patients’ clinical and molecular data
UPN
Sex
Age (yrs) Normalizeda % BM CBFMYH11 blasts at transcript diagnosis copy No. in diagnostic BM
WBC ×109/l
idarubicin, cytarabine and etoposide (ICE1 and ICE2; idarubicin 12 mg/m2 on days 1, 3 and 5, cytarabine 100 mg/m2 days 1–7, etoposide 100 mg/m2 days 1–3), and post-remission therapy with two cycles of high-dose cytarabine and mitoxantrone (HAM1 and HAM2; cytarabine 3 g/m2 every 12 h on days 1– 3, mitoxantrone 12 mg/m2 days 2–3).16 One patient (UPN 3) underwent allogeneic bone marrow transplantation (BMT) as an intensification treatment. This patient was mistakenly assigned to the group of patients with normal karyotypes; inv(16) was diagnosed retrospectively by cytogenetics. BM samples used for the current analysis were collected at diagnosis, at different time-points during CR and at relapse.
Sample procurement, RNA extraction and reverse transcription All samples were received fresh (overnight) at a central laboratory. Mononuclear cells (MNC) were separated by Ficoll gradient centrifugation and were first stored at −80°C for 2 to
LDH Relapse Normalized % BM Treatment in (normal CBF/MYH11 blasts at 2nd CR ⬍200 U/l) transcript relapse copy No. in relapse BM
UPN 1
female
41
NA
90%
49.0
596
no
—
—
UPN 2
female
57
1175
97%
8.5
629
no
—
—
UPN 3b female
46
1230
90%
36.0
381
no
—
—
UPN 4
male
26
3388
90%
48.0
600
no
—
—
UPN 5
male
18
4572
NA
26.2
330
no
—
—
UPN 6
female
35
7943
77%
18.7
349
no
—
—
UPN 7
male
55
16 596
35%
9.8
246
no
—
—
UPN 8
male
58
3715
90%
39.4
488
yes
NA
20%
UPN 9
male
56
NA
70%
3.0
265
yes
10 233
8%
UPN 10 female
49
5888
90%
68.3
716
yes
25 119
20%
UPN 11 female
33
6309
90%
170.0
1235
yes
12 303
NA
UPN 12
38
6456
90%
114.0
1098
yes
NA
70%
UPN 13 female
46
9120
74%
157.0
1297
yes
676
30%
UPN 14 female
49
10 715
NA
41.5
622
yes
1096
NA
UPN 15 female
37
17 378
80%
10.1
378
yes
20 893
NA
UPN 16 female
31
56 234
90%
35.1
568
yes
NA
15%
male
1073
Disease status
Alive (52.5)c in 1st CR — Alive (35.4) in 1st CR — Alive (74.8) in 1st CR — Alive (45.3) in 1st CR — Alive (30.2) in 1st CR — Alive (24.0) in 1st CR — Alive (43.2) in 1st CR Chemotherapy Alive (45.0) in 2nd CR Chemotherapy Alive (49.9) in 2nd CR Chemotherapy Dead (35.2) in relapse AlloBMT Alive (43.1) in 2nd CR APSCT Alive (61.5) in 2nd CR AlloBMT Alive (43.5) in 2nd CR AlloBMT Alive (28.9) in 2nd CR AlloBMT Dead (17.9) in 2nd CR APSCT Dead (42.6) in relapse —
UPN, unique patient number; BM, bone marrow; WBC, white blood count; CR, complete remission; NA, not available; AlloBMT, allogeneic bone marrow transplantation; APSCT, autologous peripheral stem cell transplantation. CBF/MYH11 copies a × 106. Normalized CBF/MYH11 copy number defined as 18S copies b UPN 3 received alloBMT in 1st CR. c ( ), months from diagnosis, ie survival in months. Leukemia
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
1074
3 days and subsequently frozen at −196°C. Total RNA from BM samples, thawed at room temperature, was extracted using the Ambion RNAcqueous kit according to the manufacturer’s directions (Ambion, Austin, TX, USA). Following quantification by spectrophotometry, the RNA was aliquoted and stored at −80°C until further use. The reverse transcription step was carried out as previously described.15
Production of cDNA standards A CBF/MYH11 fusion transcript type A was amplified by single round RT-PCR from a CBF/MYH11-positive AML patient using previously published primers and conditions.10 The amplified product was purified using a Qiagen PCR purification kit (Qiagen, Chatsworth, CA, USA), and then cloned into the pCR 2.1 vector (Invitrogen, San Diego, CA, USA) according to the manufacturer’s directions. A plasmid of 苲4300 kb was purified using a Qiagen mini-prep kit (Qiagen), according to the manufacturer’s directions. This plasmid was then diluted in DNase and RNase-free H2O (Sigma, St Louis, MO, USA) to obtain a set of standards ranging from 7 × 107 to 7 × 100 copies of CBF/MYH11 fusion transcript/l H2O. An 18S cDNA standard was constructed in a similar fashion. Briefly, 18S sequence was PCR amplified by a single round RT-PCR from human normal donor cDNA (see Table 2 for primer sequences). The amplified product was purified using a Qiagen PCR purification kit (Qiagen), and then cloned into the pCR 2.1 vector (Invitrogen) according to the manufacturer’s directions. A plasmid of 苲4000 kb was purified using a Qiagen mini-prep kit (Qiagen), according to the manufacturer’s directions. This plasmid was then diluted in DNase and RNase-free H2O (Sigma) to obtain a set of standards.
Construction of CBF/MYH11 and 18S standard curves Real Time PCR amplification of serially diluted CBF/MYH11 cDNA standards was performed using 2.5 l of the diluted cDNA, 12.5 l of 2× Master Mix (8% glycerol, 1× TaqMan buffer A, 200 m dATP, 200 m dCTP, 200 m, dGTP, 400 m dUTP, 0.05 U/l AmpErase uracil N-glycosylase, 5 mm MgCl2, 0.01 U/l Gold Amplitaq DNA polymerase (Perkin-Elmer, Foster City, CA, USA)), 2.5 l CBF/MYH11 detection probe at a final concentration of 200 nm and 5 l CBF/MYH11 forward/reverse primers at a final concentration of 900 nm. The reaction was brought up to a final volume of 25 l adding 2.5 l DNase and RNase-free H2O. Sequences of the CBF/MYH11 detection probe and primers were designed using the Primer Express program (PE Applied Biosystems, Foster City, CA, USA) and are listed in Table 2. PCR reactions were set up in a MicroAmp Optical 96-Well Table 2
Quantification of the CBF/MYH11 fusion transcript in unknown patient samples Amplifications of the CBF/MYH11 fusion transcript for the patient samples and the negative controls were performed by the Real Time PCR using 2.5 l of patient or negative control cDNA, and the same conditions described for the CBF/MYH11 cDNA standards. All the reactions for the patient and negative control samples were run in duplicate, and the ⌬Rn and CT were averaged from the values obtained in each reaction. Calculation of the absolute CBF/MYH11 fusion transcript copy number was accomplished by comparing the CT value of each patient sample to the CBF/MYH11 standard curve. In order to minimize variability in the results due to differences in the RT efficiency and/or RNA integrity among the unknown patient samples, the absolute CBF/MYH11 fusion transcript copy number was normalized to the cDNA copy number of an internal control, 18S RNA. The normalized values of the CBF/MYH11 fusion transcript copies in each patient and control sample were reported as the ratio of CBF/MYH11 fusion transcript copy number/18S transcript copy number × 106. To eliminate potential sources of variations related to transcript quantification, patients samples were run in batches soon after RNA extraction (⬍48 h). The CBF/MYH11 and 18S plasmid dilutions used for construction of the standard curves were stored at −80°C. CBF/MYH11 and 18S standard curves were accepted for quantitation only if the coefficient of correlation was ⭓0.995.
Probes and primer sets used for the Real Time PCR quantification
Real Time PCR oligonucleotide set CBF/MYH11 probe CBF/MYH11 forward primer CBF/MYH11 reverse primer 18S probe 18S forward primer 18S reverse primer
Leukemia
Reaction Plate (Perkin-Elmer) and performed on the Model 7700 Sequence Detector (PE Applied Biosystems) as previously reported.15 All the reactions for the CBF/MYH11 standards were run in triplicate, and the ⌬Rn (intensity of the target-specific normalized fluorescent signal) and CT (number of PCR cycles necessary to achieve a target-specific fluorescence detection threshold) were averaged from the values obtained in each reaction (Figure 1a). A CBF/MYH11 standard curve was then constructed plotting the CT vs the known copy number of each standard sample (coefficient of correlation 0.998) calculated at the detection threshold (Figure 1b). Similarly, a limiting dilution of the 18S plasmid was amplified by Real Time PCR using the same conditions described above (see Table 2 for primer and probe sequences). An 18S standard curve was then constructed plotting the CT vs the known copy number of each standard sample (coefficient of correlation 0.995) calculated at the detection threshold.
Sequence 5′ → 3′ FAM-CTCCATTTCCTCCCGATGAGACCTGTCT-TAMRA AGAAGGACACGCGAATTTGAA TGGACTTCTCCAGCTCATGGA JOE-TGCTGGCACCAGACTTGCCCTC-TAMRA CGGCTACCACATCCAAGGAA GCTGGAATTACCGCGGCT
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
1075
Figure 1 Construction of the CBF/MYHII standard curve. (a) Real Time amplification plots (done in triplicate) for different dilutions of a CBF/MYH11 cDNA standard, ranging between 7 × 105 CBF/MYH11 cDNA copies/l (the first plot from the left) to 7 × 100 CBF/MYH11 cDNA copies/l (the last plot from the left). The value ⌬Rn, normalized intensity of target-specific fluorescent signal detected during PCR for each dilution, is plotted against the number of PCR cycles. The intersection of each amplification plot with the detection threshold (the solid horizontal line) defined the CT (number of PCR cycles necessary to achieve a detection threshold) value for each standard dilution. (b) The CBF/MYH11 standard curve. The CT values obtained by amplifying each CBF/MYH11 standard dilution were plotted against the corresponding CBF/MYH11 copy number present in each CBF/MYH11 standard dilution before the PCR amplification. The CT decreased linearly with the increase of the CBF/MYH11 cDNA copy number, since the higher the CBF/MYH11 copy number initially present in a standard dilution, the lower the PCR cycle number necessary to achieve the detection threshold. A standard curve was thus generated and used to calculate the CBF/MYH11 fusion transcript copy number in patient samples.
Sensitivity of Real Time PCR The sensitivity of the Real Time PCR assay was determined by serial dilutions of CBF/MYH11-positive RNA in CBF/MYH11-negative RNA. The RNA sample from the diagnostic BM of UPN 12 (90% blasts; 6456 CBF/MYH11 fusion transcript copies), was serially diluted with CBF/MYH11negative RNA. Two g of the RNA mixtures were then reverse-transcribed and Real Time-PCR amplified using the conditions described above. The sensitivity for amplification of the CBF/MYH11 fusion transcript by Real Time RT-PCR was found to be 1:104 (Figure 2a and b).
Statistical analysis Two of the 16 patients included in this study did not have a diagnostic BM sample available for analysis, and one patient underwent allogeneic BMT as intensification treatment in first CR. Therefore, inferential statistics assessing the impact of baseline covariates on clinical outcome were calculated using data from the remaining 13 patients. Descriptive statistics such as the median, minimum, and maximum were reported for continuous variables such as age, CBF/MYH11 fusion transcript levels, BM blast percentages, WBC and LDH values.
Dichotomous variables such as gender were summarized by reporting the total and percent. Due to the small sample size and non-adherence of the data to normality assumptions, nonparametric statistics were used in the hypothesis testing procedures. To assess the relation between pairs of variables such as diagnostic CBF/MYH11 fusion transcript level, LDH, WBC, BM blasts, CR duration, survival, and age, Spearman’s rank correlation coefficients and the corresponding P values were reported. The Wilcoxon rank sum test was used to assess whether there were any significant differences in diagnostic CBF/MYH11 fusion transcript level, WBC and age by gender (male vs female) and clinical outcome (CCR vs relapse). All tests were two-sided and results were considered significant at the 0.05 level. To further investigate whether the CBF/MYH11 fusion transcript level was predictive of clinical outcome after completion of the entire treatment program, the fusion transcript copy number detected in the last follow-up sample during remission was used. This variable was dichotomized into ‘High’ (copy number ⬎10; n = 5) vs ‘Low’ (copy number ⬍10; n = 5) by examining the distribution of the 10 patients with such a measure and breaking it at a natural cutpoint (10 copies). A log-rank test was performed to assess whether the ‘High’ and ‘Low’ groups differed with respect to CR duration. Furthermore, to account for the time-dependent nature of this Leukemia
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
1076
Figure 2 Sensitivity of the CBF/MYH11 Real Time RT-PCR assay. Serial dilutions of CBF/MYH11-positive RNA from a diagnostic BM sample (UPN 12) into CBF/MYH11-negative RNA were amplified by Real Time RT-PCR. The sensitivity (a; ie, 1:104) and the linearity (b; correlation coefficient 0.987) of the assay are shown. U, undiluted sample.
last copy number assessment, a log-rank test was performed to assess whether the ‘High’ and ‘Low’ groups differed with respect to CR duration and survival from this last assessment. Corresponding Kaplan–Meier curves were constructed for CR duration from the last copy number assessment prior to relapse/censoring. Finally, Fisher’s exact test was used to compare the clinical outcome defined as relapse vs CCR for the ‘High’ vs ‘Low’ groups. Fisher’s exact test was also used to compare the clinical outcome defined as relapse vs CCR for two groups of patients, those included in the current analysis (n = 16) and those excluded on the basis of lack of sample availability (n = 7). Additionally, a log-rank test was performed to determine whether overall survival differed between the included and excluded groups. Results
Clinical outcome A total of 56 diagnostic, follow-up and relapse BM samples from 16 patients with AML and inv(16)(p13q22) were retrospectively analyzed for levels of the CBF/MYH11 fusion transcripts using the Real Time RT-PCR (Table 3). All patients included in this analysis had a type A CBF/MYH11 fusion transcript and achieved CR following induction chemotherapy. For the 16 patients included in this study, the median CR duration was 19 months (range 6.5–73.8 months). The median survival was not reached, since of the 16 patients, only three have died (current median survival among living patients 43.5 months; range 24.0–74.8 months). Nine of the Leukemia
16 patients relapsed (median time to relapse 11.5 months; range 6.5–21.6 months), while seven patients are in CCR (current median CR duration 42.1 months; range 23–73.8 months). All nine patients who relapsed entered second CR following re-induction chemotherapy. In six patients re-induction was followed by high-dose chemotherapy with allogeneic bone marrow transplantation (alloBMT) (n = 4; UPNs 11, 13, 14, 15) or autologous peripheral stem cell transplantation (APSCT) with stem cells collected in first CR (n = 2; UPNs 12 and 16). Three patients (UPNs 8, 9, 10) received salvage chemotherapy. Of the nine patients who relapsed, six (67%) (UPNs 8, 9, 11, 12, 13, 14) are alive and in CR. Thus, of the 16, 13 patients (81%) are alive, in first or second CR. Seven additional patients with inv(16) were enrolled on the AML HD93 protocol and received the same induction and intensification treatment, but were not included in the current study because of lack of suitable samples for the molecular analysis. Of these seven patients, all achieved CR; subsequently one relapsed and six continue in CR. There was no statistically significant difference between the patients included (n = 16) and those excluded (n = 7) from the molecular analysis with respect to clinical outcome defined as relapse vs CCR (P = 0.09; Fisher’s exact test). Furthermore, there was no significant difference in survival duration comparing patients included and those excluded (P = 0.81; log-rank test).
Quantification of CBF/MYH11 expression in diagnostic samples Fourteen diagnostic BM samples were available for analysis. First, we determined that the Real Time RT-PCR could be used
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
Table 3
Patient
UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN UPN
1 2 3b 4 5 6 7 8 9 10 11 12 13 14 15 16
1077
Quantitation of the CBF/MYH11 fusion transcript over time
Diagnosis
p ICE 1
NA 1175c 1230 3388 4572 7943 16 596 3715 NA 5888 6309 6456 9120 10 715 17 378 56 234
119 (0)d
p ICE 2
p HAM 1
p HAM 2
6 (11.8)a 7 (11.1)a 1 (24.7)
4 (2.4) 8 (0)
5 (2.3) 1 (2.7)
Remission f/u No. 1 (first follow-up from completion of tx)
Remission f/u Remission No. 2 f/u No. 3
41 (32.5)
Relapse
0 (66.5)
2 (3.9) 3 (4.1) 1 (5.0)a 7 (25.8)a
589 (0) 3 (0) 1694 (0)
3 (3.2) 31 (1.1)
2 (4.6) 5 (2.7) 17 (2.4)
2 (3.2) 8 (4.0)
17 (3.3) 6 (2.4) 2 (3.1) 7 (3.0)
5 (11.7)a 24 684 (9.6)a 250 59 58 8
(7.0) (9.9)a (5.3)a (5.1)
19 (8.1)a 133 (8.9)a
NA (16.4) 10 233 (21.6) 25 118 (11.5) 12 303 (7.5) NA (11.5) 676 (12.4) 1096 (6.5) 20 893 (11.3) NA (10.3)
p, post; f/u, follow-up; tx, treatment; NA, not available. a Sample used to assess the prognostic value of the CBF/MYH11 fusion transcript copy number detected at the last remission f/u. b Last BM sample from UPN 3 was not included in the prognostic analysis done during CR since this patient received alloBMT during 1st CR. c Normalized CBF/MYH11 fusion transcript copy number. d (n), months from achievement of 1st CR.
to detect variation in the fusion transcript copy number at different time-points during the course of the disease. As expected, the highest fusion transcript copy numbers were detected at diagnosis and relapse, whereas the copy number decreased following initial and salvage chemotherapy, and increased again just before clinical relapse (see representative case in Figure 3). Second, we assessed whether the CBF/MYH11 fusion transcript copy number in diagnostic BM samples had any prognostic value in predicting CCR in this homogeneously treated patient population. We excluded from this analysis UPN 3, a patient treated with allogeneic BMT in first CR. The CBF/MYH11 fusion transcript copy number in the remaining
13 diagnostic BM samples ranged between 1175 and 56 234 copies per 106 copies of 18S, with a median value of 6456. At diagnosis, the median CBF/MYH11 fusion transcript copy number in the eight patients who relapsed and have a diagnostic BM sample available for analysis was 7788 (range 3715–56 234), whereas the median was 4572 (range 1175– 16 596) in the five patients who remain in CCR (Table 1). Using a Wilcoxon rank-sum test, we found that, at diagnosis, clinical outcome defined as CCR or relapse was adversely impacted by a higher WBC (P = 0.05), but not by a higher CBF/MYH11 copy number (P = 0.21) or older age (P = 0.61). Using the Spearman’s rank correlation, CBF/MYH11 fusion transcript copy number at diagnosis was correlated with age, WBC, hemoglobin, platelets, LDH, BM blasts, CR and survival duration. A significant correlation was found between a high CBF/MYH11 fusion transcript copy number and a high percentage of BM blasts (Spearman’s correlation = −0.66; P = 0.03). A borderline significant correlation was also found between a high CBF/MYH11 fusion transcript copy number and a short CR duration (Spearman’s correlation = −0.51; P = 0.07) (Figure 4).
Quantification of CBF/MYH11 in follow-up samples
Figure 3 Quantification of the CBF/MYH11 transcript copy number over time. The graph shows variation in the CBF/MYH11 transcript level over time (exemplified using values from patient UPN 16). The highest CBF/MYH11 transcript copy numbers were detected at diagnosis and relapse, whereas the copy number decreased following initial and salvage treatments, and increased again just before clinical relapse occurred. ICE and HAM (see text); HiDAC, high-dose cytarabine; APSCT, autologous peripheral stem cell transplantation.
All patients achieved CR following induction treatment (ICE 1 and 2). A reduction in the CBF/MYH11 fusion transcript copy number was detected in remission samples collected during and following completion of treatment with respect to the diagnostic samples (Table 3). No difference in the CBF/MYH11 fusion transcript copy number was found during intensification treatment among patients destined to remain in CCR compared to patients destined to relapse (P = 0.75; Wilcoxon rank sum test). To assess the value of CBF/MYH11 fusion transcript quantification in predicting clinical outcome, we calculated Leukemia
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
1078
Figure 4 Plot of CBF/MYH11 transcript copy numbers at diagnosis vs CR duration in months. The closed symbols represent the relapsed patients, whereas the open symbols represent patients in CCR.
the copy numbers in last remission BM samples procured following completion of the entire treatment program. Ten of the 15 patients who received chemotherapy as intensification had a BM sample available for this analysis (Table 3). We identified two groups of patients, one with a CBF/MYH11 fusion transcript copy number ⬍10 (‘Low’ group) and one with a CBF/MYH11 fusion transcript copy number ⬎10 (‘High’ group). These two groups were compared for risk of relapse and CR duration. In the ‘Low’ group, the CBF/MYH11 fusion transcript copy number was assessed at a median time of 11.7 months from achievement of CR (range 5.0–25.8 months); in the ‘High’ group, this value was assessed at a median time of 8.9 months from achievement of CR (range 5.3–9.9 months). There was no significant difference between these groups with respect to the sampling time (P = 0.14; Wilcoxon rank sum test). Patients in the ‘High’ group had a significantly higher risk of relapse than patients in the ‘Low’ group (P = 0.05; Fisher’s exact test). A significantly shorter CR duration was also found between the ‘High’ and ‘Low’ group (P = 0.002; log-rank test). Additionally, after accounting for the time-dependent nature of the sampling, a significant difference in CR duration calculated from the time of the last sample assessment was found between the ‘High’ and ‘Low’ group (P = 0.002; log-rank test). The median CR duration from the last sampling for the ‘High’ group was 2.3 months (range 1.2–3.5 months), whereas the median CR duration from last sampling for the ‘Low’ group could not be estimated since ⬍50% of the patients in this group relapsed (only one patient relapsed at 9.9 months from the last sampling) (Figure 5). There was not a significant difference in survival duration from the last copy number assessment between the ‘High’ and ‘Low’ group (P = 0.2; log-rank test). Two patients have died in the group with ⬎10 CBF/MYH11 transcript copies, whereas no patient has died in the group with ⬍10 CBF/MYH11 transcript copies (see data in Tables 1 and 3). Discussion Although patients with inv(16)(p13q22) AML have a relatively good prognosis when optimally treated with induction and Leukemia
Figure 5 Kaplan–Meier plot of CR duration from last copy number assessment according to whether patients had a CBF/MYH11 transcript copy number of more or less than 10 copies in the last followup remission BM sample.
HiDAC intensification chemotherapy, a significant number of them relapse and eventually die of their disease. Early identification of those patients at high risk for unfavorable clinical outcome has been hypothesized to be important to design diversified and more aggressive therapeutic strategies tailored to the probability of disease relapse. Several groups have explored the use of CBF/MYH11 fusion transcript amplification by RT-PCR for detection of MRD and prediction of relapse in inv(16)(p13q22) AML patients, but these studies have, to date, produced conflicting results.12 Costello et al11 have recently suggested that achievement of RT-PCR negativity within 8 months from induction chemotherapy in inv(16)(p13q22) AML may be predictive of continuous remission, whereas other groups have previously reported patients in long-term CCR, despite never becoming negative for the CBF/MYH11 fusion transcript by RT-PCR.7,8 The reasons for these conflicting results are probably related to lack of standardized experimental conditions, the heterogeneity of the sample collections (eg BM vs blood vs leukapheresis products), differences in the sampling time-points and in the intensity of the received treatment (eg conventional vs high-dose dose chemotherapy; alloBMT vs APSCT). Thus, although a total of 37 patients with CBF/MYH11 AML have been analyzed for MRD in seven different studies, it is not possible to predict disease relapse or cure only on the basis of a positive or negative RT-PCR.6–12 Two recent small retrospective studies have suggested that a quantitative RT-PCR methodology that detects residual levels of CBF/MYH11 fusion transcripts in inv(16)(p13q22) AML during CR may prove more useful to assess treatment response and predict clinical outcome than a simple qualitative assay.13,14 Both these studies, using a quantitative competitive RT-PCR (QcRT-PCR), found that the decline in the amount of fusion transcript following treatment occurs over a variable length of time, and suggested that the degree of log-fold reduction in the level of the CBF/MYH11 fusion transcript is predictive of patients’ outcome. In neither study, however, was it possible to identify, during CR, an absolute threshold of fusion transcripts above which relapse occurs and below which cure is likely. Moreover, these studies did not address whether the level of the CBF/MYH11 fusion transcript at diagnosis has any value in predicting clinical outcome. The results of these two studies are certainly intriguing, but
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
limitations such as the small number of patients analyzed (n = 7 and n = 5, respectively) and the narrow dynamic range of quantification of the technique utilized prohibit any firm conclusion from being drawn. The recent introduction of the Real Time RT-PCR appears promising for resolving some of the technical limitations intrinsic in these initial quantitative analyses. The Real Time PCR (or Taqman PCR) is a fluorogenic- based PCR methodology that allows collection of amplification data in real time (ie during the PCR reaction) for a given target using an automated technology.15,17 Although a recent comparison of Real Time RT-PCR with QcRT-PCR using samples from t(8;21) AML patients showed that these two methodologies are comparable for sensitivity, linearity and reproducibility, the former method of analysis appears to offer technical advantages by providing absolute quantification of the target sequence, expanding the dynamic range of quantification to over six orders of magnitude, eliminating the post-PCR processing, and reducing labor and carryover contamination.16 In the current study, we have used Real Time RT-PCR to quantify the CBF/MYH11 fusion transcript in diagnostic, remission and relapse samples from 16 patients with inv(16)(p13q22) AML uniformly treated on the German multicenter study AML HD93. To our knowledge, this is the largest patient series of quantitative RT-PCR analyses reported for inv(16)(p13q22) AML. Although the retrospective nature and the small sample size of the current study are limiting factors, interesting points emerged from the analysis of these results. First, our data suggest that the CBF/MYH11 fusion transcript copy number detected in remission BM samples obtained following completion of the entire therapeutic program is a predictor of subsequent clinical outcome (Figure 5). We calculated the CBF/MYH11 fusion transcript copies in BM samples available at the longest remission time-point. All the patients considered for this analysis, including UPN 10 who showed at this time-point a very high CBF/MYH11 fusion transcript copy number, were deemed in hematologic CR on the basis of morphologic assessment of the BM samples (⬍5% blasts; no evidence of leukemic cells). Karyotype analysis of these remission samples was not performed and, therefore, the possibility of cytogenetic relapse at the time of BM sampling cannot be excluded. However, none of the remission samples were collected in response to specific changes in blood counts suggestive of impending disease relapse. Thus, extreme variation in the CBF/MYH11 fusion transcript copy number as seen in UPN 10 could be due to a technical error, or illustrate that expression of a chimeric transcript in a residual leukemic clone may vary dramatically from patient to patient. We found, however, that patients with ⬎10 CBF/MYH11 fusion transcript copies had a shorter CR duration (P = 0.002) and higher risk of relapse (P = 0.05) than patients with ⬍10 copies. This analysis was based on evaluation of remission BM samples procured at a median time of 8.9 months from achievement of CR for patients with ⬎10 copies and at a median time of 11.7 months from achievement of CR for patients with ⬍10 copies. The difference in sampling time between patients with ⬎10 (‘High’ group) or ⬍10 copies (‘Low’ group) was not significant (P = 0.14). In order to correct for the sampling time variability, however, we compared the two groups for CR duration from the time of the sample procurement. We found that the ‘High’ group had a significantly shorter CR duration from the last assessment of copy number during remission than the ‘Low’ group (P = 0.002). It is unlikely, therefore, that the longer CR duration in the group with ⬍10 copies is a mere reflection of a risk of relapse that
decreased as a function of passage of time. These results suggest that it may ultimately be possible to determine during remission a threshold of MRD or CBF/MYH11 expression predictive of CCR or relapse. This conclusion is further supported by comparing the CCR duration in the ‘Low’ group with the time to relapse in the ‘High’ group. At the last followup, four patients in the ‘Low’ group continued in CCR with a minimum duration of remission of 23 months, 1.4 months beyond the longest time to relapse (ie 21.6 months) observed in the ‘High’ group where all five patients have relapsed. A prognostic significance of chimeric gene expression during CCR has also been reported in another recent study. Tobal et al18 evaluated 21 patients with t(8;21)(q22;q22) AML for MRD by QcRT-PCR. As in our study, comparison of levels of the AML1/ETO fusion transcript detected in remission samples from CCR patients with those detected in remission samples from relapsed patients enabled the establishment of a threshold for MRD predictive of impending hematologic relapse. Second, we found no difference in the CBF/MYH11 fusion transcript copy number calculated during intensification chemotherapy among patients destined to relapse and those destined to continue in CR (P = 0.75), suggesting that early quantification of chimeric fusion transcripts may not be a feasible strategy to predict treatment response and final clinical outcome. Since it is becoming clear from recent studies with transgenic mice that expression of CBF/MYH11 is not a sufficient condition for developing leukemia, but additional genomic ‘hits’ are necessary, it would be interesting to use developing methodologies (eg microarray gene expression analyses) to characterize genomic abnormalities that occur simultaneously with the CBF/MYH11 fusion.19,20 The ultimate goal of this approach should be identification of those genomic aberrations in CBF/MYH11 leukemic blasts whose presence at diagnosis or persistence during treatment would allow early prognostic stratification without need for quantification of CBF/MYH11 expression. Finally, in our study, patients with higher diagnostic CBF/MYH11 transcript copy numbers appear to have a shorter CR duration, but such correlation did not achieve statistical significance (P = 0.07) (Figure 4). Although intriguing, the findings of our investigation require confirmation. Future large prospective studies that analyze cohorts of homogeneously treated patients and utilize BM and/or PB samples collected at the same specific time-points are necessary to establish the prognostic value of fusion transcript quantification in CBF/MYH11 inv(16)(p13q22) AML.
1079
Acknowledgements This work was supported in part by grant P30CA16058, National Cancer Institute, Bethesda, MD, The Coleman Leukemia Research Foundation and The American Cancer Society IRG-98–278–01 grant.
References 1 Mro´zek K, Bloomfield CD. Chromosome aberrations in de novo acute myeloid leukemia in adults: clinical implications. Rev Clin Exp Hematol 1998; 5: 44–67. 2 Bloomfield CD, Lawrence D, Byrd JC, Carroll A, Pettenati MJ, Tantravahi R, Patil SR, Davey FR, Berg DT, Schiffer CA, Arthur DC, Mayer RJ. Frequency of prolonged remission duration following Leukemia
Quantification of CBF/MYH11 fusion transcript by Real Time RT-PCR G Marcucci et al
1080 3 4
5 6
7
8
9
10
11
12
Leukemia
high-dose cytarabine intensification in acute myeloid leukemia varies by cytogenetic subtype. Cancer Res 1998; 58: 4173–4179. Yin JA, Tobal K. Detection of minimal residual disease in acute myeloid leukaemia: methodologies, clinical and biological significance. Br J Haematol 1999; 106: 578–590. Liu P, Tarle SA, Hajra A, Claxton DF, Marlton P, Freedman M, Siciliano MJ, Collins FS. Fusion between transcription factor CBF/PEBP2b and a myosin heavy chain in acute myeloid leukemia. Science 1993; 261: 1041–1044. Liu PP, Hajra A, Wijmenga C, Collins FS. Molecular pathogenesis of the chromosome 16 inversion in the M4Eo subtype of acute myeloid leukemia. Blood 1995; 85: 2289–2302. Claxton DF, Liu P, Hsu HB, Marlton P, Hester J, Collins F, Deisseroth AB, Rowley JD, Siciliano MJ. Detection of fusion transcripts generated by the inversion 16 chromosome in acute myelogenous leukemia. Blood 1994; 83: 1750–1756. Hebert J, Cayuela JM, Daniel MT, Berger R, Sigaux F. Detection of minimal residual disease in acute myelomonocytic leukemia with abnormal marrow eosinophils by nested polymerase chain reaction with allele specific amplification. Blood 1994; 84: 2291–2296. Tobal K, Johnson PR, Saunders MJ, Harrison CJ, Liu Yin JA. Detection of CBF/MYH11 transcripts in patients with inversion and other abnormalities of chromosome 16 at presentation and remission. Br J Haematol 1995; 91: 104–108. Poirel H, Radford-Weiss I, Rack K, Troussard X, Veil A, Valensi F, Picard F, Guesnu M, Leboeuf D, Melle J. Detection of the chromosome 16 CBF beta-MYH11 fusion transcript in myelomonocytic leukemias. Blood 1995; 85: 1313–1322. Elmaagacli AH, Beelen DW, Kroll M, Trzensky S, Stein C, Schaefer UW. Detection of CBFbeta/MYH11 fusion transcript in patients with inv(16) acute myeloid leukemia after allogeneic bone marrow or peripheral blood progenitor cell transplantation. Bone Marrow Transplant 1998; 21: 159–166. Costello R, Sainty D, Blaise D, Gastaut JA, Gabert J, Poirel H, Buzyn-Veil A, Macintyre E. Prognosis value of residual disease monitoring by polymerase chain reaction in patients with CBFbeta/MYH11-positive acute myeloblastic leukemia. Blood 1997; 89: 2222–2223. Marcucci G, Caligiuri MA, Bloomfield CD. Defining the ‘absence’
13
14
15
16
17 18
19
20
of the CBF/MYH11 fusion transcript in patients with acute myeloid leukemia and inversion of chromosome 16 to predict long-term complete remission: a call for definitions. Blood 1997; 90: 5022–5024. Evans PA, Short MA, Jack AS, Norfolk DR, Child JA, Shiach CR, Davies F, Tobal K, Liu Yin JA, Morgan GJ. Detection and quantitation of the CBFbeta/MYH11 transcripts associated with the inv(16) in presentation and follow-up samples from patients with AML. Leukemia 1997; 11: 364–369. Laczika K, Novak M, Hilgarth B, Mitterbauer M, Mitterbauer G, Scheidel-Petrovic A, Scholten C, Thalhammer-Scherrer R, Brugger S, Keil F, Schwarzinger I, Haas OA, Lechner K, Jaeger U. Competitive CBFbeta/MYH11 reverse-transcription polymerase chain reaction for quantitative assessment of minimal residual disease during postremission therapy in acute myeloid leukemia with inversion(16): a pilot study. J Clin Oncol 1998; 16: 1519–1525. Marcucci G, Livak KJ, Bi WL, Strout MP, Bloomfield CD, Caligiuri MA. Detection of the AML1/ETO fusion transcript in patients with t(8;21)-associated AML using a novel ‘real time’ quantitative RTPCR assay. Leukemia 1998; 12: 1482–1489. Wattjes MP, Krauter J, Nagel S, Heidenreich O, Ganser A, Heil G. Comparison of nested competitive RT-PCR and real-time RTPCR for the detection and quantification of AML1/MTG8 fusion transcripts in t(8;21) positive acute myelogenous leukemia. Leukemia 2000; 14: 329–335. Heid CA, Stevens J, Livak KJ, Williams PM. Real time quantitative PCR. Genome Res 1996; 6: 986–994. Tobal K, Newton J, Macheta M, Chang J, Morgenstern G, Evans PA, Morgan G, Lucas GS, Liu Yin JA. Molecular quantitation of minimal residual disease in acute myeloid leukemia with t(8;21) can identify patients in durable remission and predict clinical relapse. Blood 2000; 95: 815–819. Castilla LH, Garrett L, Adya N, Orlic D, Dutra A, Anderson S, Owens J, Eckhaus M, Bodine D, Liu PP. The fusion gene CbfbMYH11 blocks myeloid differentiation and predisposes mice to acute myelomonocytic leukaemia. Nat Genet 1999; 2: 144–146. Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, Bloomfield CD, Lander ES. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 1999; 286: 531–537.
